GenAI as an Accessibility Tool

Sat Feb 28 2026

There’s a lot of popular backlash against Generative AI. I’m aiming in this piece to give my thoughts about AI, as someone who has studied it formally at JHU, at least the basic principles and implementation, and my relationship with it.

I can put it simply; AI to replace humans is a bad idea. In music and art, I think we are just inevitably going to see a rise of AI generated pieces. I honestly believe there’s a place for that - as AI music, AI art. Not human music and art. But companies seeking to eradicate human workers in favor of AI is not the move, in my opinion, if for no other reason than the market is going to tank when everyone is out of work and losing housing, as we’re starting to see more of in the US.


I believe it is a better idea to use AI to superpower humans, so to speak; to accelerate us and aid us. Science, medicine, forms of research are all greatly enhanced by various applications of AI. Most of what I want to bring light to today is how I use GenAI to overcome cognitive barriers I have from probable neurodivergence and long term mental illness.



I’ve always been a thoughtful engineer. I’ve always wanted to know the “why” of the code, wanted to do things to the highest quality possible. As I moved past more rudimentary code in my education, I began to struggle with working memory. The nodes of logic I could have stored in my quick access memory were few, I started to lose them after even 2-3 some days. Anxiety and panic disorder play a role, which makes the mechanical difficulty even worse.


I struggle to interview as an engineer. I am TERRIBLE at technical interviews. Because of these working memory issues, usual anxiety and the additional anxiety of the interview, I just cannot perform well. I understand the concepts and will randomly come up with the architecture later that night during dinner, but on the spot it’s a no go. I am a slower thinker and this world is much more interested in fast than thorough.

But, with the advent of Gen AI aiding engineers, the tools we’re turning to are also a disability aid for me. Instead of getting in the muck of the execution constantly, I can come up with a game plan, give my AI buddy the plan and code review it as it comes up with implementation bits.

I am probably of a dying breed of people who know how to engineer the web by hand. I am from a time where we didn’t have console warnings or even browser inspection tools. Now, I also embrace a healthy amount of AI assistance. I feel I can do that to good effect because I know what the AI should ideally be generating. I can catch it in mistakes and things it overlooks. I can endlessly remind it that it must create database migrations, revalidate types while working in Payload CMS, that we are indeed writing strict mode TypeScript and that it should write automated tests. I know to do this, where others may not. So, I do argue that my usage is fairly ideal in terms of application and execution.

What do you think? Are there ways you’re brave enough to admit to using UI in your workflow to great effect? Perhaps even overcoming your own difficulties?

← Back to blog